| Summary: | pxlcolor: differences in output | ||
|---|---|---|---|
| Product: | Ghostscript | Reporter: | Marcos H. Woehrmann <marcos.woehrmann> |
| Component: | PXL Driver | Assignee: | Hin-Tak Leung <htl10> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | Keywords: | bountiable |
| Priority: | P4 | ||
| Version: | master | ||
| Hardware: | Macintosh | ||
| OS: | MacOS X | ||
| Customer: | Word Size: | --- | |
| Attachments: |
bigpaths.ps
bigpaths.ps.png colored_pattern.ps colored_pattern.ps.png uncolored_pattern.ps uncolored_pattern.ps.png 5570 pxl streams clip to max_16bit clip signed shorts and signed short pairs to (-32768, 32767) |
||
|
Description
Marcos H. Woehrmann
2009-10-28 19:31:10 UTC
Created attachment 5570 [details]
bigpaths.ps
Created attachment 5571 [details]
bigpaths.ps.png
Created attachment 5572 [details]
colored_pattern.ps
Created attachment 5573 [details]
colored_pattern.ps.png
Created attachment 5574 [details]
uncolored_pattern.ps
Created attachment 5575 [details]
uncolored_pattern.ps.png
Created attachment 5576 [details]
5570 pxl streams clip to max_16bit
Actually both of 5571 are wrong - the correct appearance should be 4 different
colors to 4 corners of the box. The number oveflows differently on different gs
devices. (and different platform also).
This is a best attempt (clip the numbers to largest representable), but pcl6
appears to do it wrongly. Does a printer do this pxl stream the same way as
pcl6?
Created attachment 5577 [details]
clip signed shorts and signed short pairs to (-32768, 32767)
clip signed shorts and signed short pairs to (-32768, 32767). The fixes the
problem with 5570 somewhat so it is closer than how it should be (the direct
ppmraw result in 5571 is actually wrong) but the cursor co-ordinates in PXL is
limited to either signed short pairs or short pairs, (i.e. both (-32768, 32767)
or both (0, 65535).
Maybe one can massage the CTM temporarily to turn a large number into within
limit, and massage it back for such ridiculously big paths. need a little time
to think.
Comment on attachment 5577 [details] clip signed shorts and signed short pairs to (-32768, 32767) part of the big patch, attachment 5599 [details] (Bug 690858) addresses set cursor 16-bit overflows much more satistfactorily. bigpath.ps should work after that patch. uncolored_pattern.ps is fixed by the patch in 690830 plus the corrected patch in 688372. the problem with uncolored_pattern.ps is essentially the same as 690025 - it has got both white-on-transparent and black-on-transparent, although one is a star shape and the other is a font glyph. colored_pattern.ps, although looks completely different, is fixed by the same patch as bug 690862 - to implement data_bit properly in write_image_data(). Argh, the three files has three different sets of problems and basically cover all the problems recently fixed in the pxl driver. bigpaths.ps is a set_cursor co-ordinate overflow, and fixed by r10271 'Use SetPageScale to work around 16-bit limit in SetCursor' (and r10270 'new SetPageScale/Unset utility routine'). It also have negative co-ordiates so r10267 'write signed 16-bit integers' is also relevant). colored_pattern.ps is fixed by r10313, 'handle data_bit'. uncolored_pattern.ps is a white-on-mask and black-on-mask case and fixed by r10305 (and r10303). |